Visual Localization on Tokyo-LoD Sequence-Traj. v3
50.3Accuracy (2m-2°)LoD-Loc v3c
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LoD-Loc v3cRefinement Mode=Full, Weighting=confidence-based2026.03 | 50.3 | 79.9 | 97.3 | 1.98 | 0.23 | |
| LoD-Loc v3aRefinement Mode=Full, Weighting=area-based2026.03 | 49.8 | 79.9 | 95.8 | 2 | 0.23 | |
| LoD-Loc v3aRefinement Mode=no select, Weighting=area-based2026.03 | 42.3 | 71.7 | 92.7 | 2.21 | 0.27 | |
| LoD-Loc v3cRefinement Mode=no select, Weighting=confidence-based2026.03 | 41.6 | 69.1 | 89.8 | 2.24 | 0.28 | |
| LoD-Loc v3cRefinement Mode=no refine, Weighting=confidence-based2026.03 | 18.6 | 38.2 | 75.6 | 3.58 | 0.72 | |
| LoD-Loc v3aRefinement Mode=no refine, Weighting=area-based2026.03 | 18.6 | 38.2 | 75.6 | 3.58 | 0.72 | |
| LoD-Loc v2Refinement Mode=no refine2026.03 | 9 | 23.1 | 57.4 | 4.56 | 0.91 | |
| MC-LocBackbone=RoMa2026.03 | 3.58 | 10.72 | 31.12 | 5.16 | 0.48 | |
| MC-LocBackbone=DINOv22026.03 | 3.42 | 10.5 | 29.81 | 5.16 | 0.49 | |
| LoD-Loc v2Refinement Mode=no select2026.03 | 2.9 | 7.3 | 24.6 | 8.53 | 1.43 | |
| LoD-Loc v2Refinement Mode=Full2026.03 | 2.3 | 6.8 | 24 | 8.75 | 1.52 | |
| Prior2026.03 | 0.3 | 0.8 | 8.8 | 8.19 | 1.76 | |
| CAD-Loc2026.03 | 0 | 0 | 0 | — | — | |
| LoD-Loc2026.03 | 0 | 0 | 5.06 | 8.19 | 1.76 |